Analytics
Measure actions against a holdout
The test-store preview compares randomized groups instead of crediting every post-action purchase to sellwise.
Eligible test-store visitors can now be assigned to Treatment or Holdout before an action is applied. Holdout receives no tested action and incurs no committed incentive cost.
Results separate assignment, exposure, purchases and cost, with outcome windows and uncertainty shown explicitly. Small samples remain “Still learning.” These development scenarios validate measurement behavior, not production uplift; production merchant activation and automatic optimization remain unavailable.